Hi Alan,

this seems really nice, thanks.

The average is, as you say, the same as doing a naive weighted average, 
the true question is the correlated systematics. I attach a file 
(yCP_results) with some breakdown of systematics, as far as I got (not 
complete, not including all results, but perhaps to be of some help 
nevertheless).

I made a quick weighted average of semileptonic results. I get
R_M = (0.50 +- 0.45) x 10-3
(neglecting any correlation). A similar file as for the y_CP is also 
attached (with the list of results being combined).

Then I tried to combine the 2-d likelihood from semileptonic average and 
the one from y_CP. These plots are attached as well:
- semil_cont.ps: contours in x and y plane from semileptonic decays 
(white color represents a 68.3% C.L., i.e. the area where 2logL 
increases by < 2.3);
- ycp_cont.ps: contours from y_CP average (white color represent the 
area with 2logL increase < 1)
- com_pl.ps: contours for combined (sum) logL (again white color 
corresponds to 2DeltaLogL < 2.3).

Of course likelihoods for y_CP and R_M are assumed to be parabolic. 
Perhaps I made some error, I certainly appreciate your opinion.

Cheers,
Bostjan.